Polyxia is a voice assistant project created by DevOps students at Polytech Montpellier that includes a FaaS architecture based on Rik and Firecracker. Rik is an experimental workload orchestrator that enables deployment of cloud applications written in Rust, providing a way to manage and schedule application instances. Firecracker, on the other hand, is an open-source project by AWS that allows for lightweight, microVM-based isolation of workloads, designed to be used with containers or serverless functions for an additional layer of security and isolation.
In addition, Polyxia features an NLU algorithm for detecting user intent from text generated through a speech-to-text algorithm that translates the user's voice command into text. This intent is then sent to the FaaS to schedule a function associated with the intent.
- FaaS architecture based on Rik and Firecracker
- NLU algorithm for intent detection from user text input
- Speech-to-text algorithm for converting user voice commands into text
- Scheduled function execution based on user intent
📒 Projects | 🔭 Status | ✏️ Description |
---|---|---|
Polyxia | Main repository for the Polyxia project, entry point for the documentation and the project management | |
Rik | Fork of the Rik project | |
Infrastructure | This repository contains files for deployment of PolyXia | |
Morty | Morty is the official CLI to manage / invoke functions over Polyxia. | |
Alpha | Alpha is a lightweight agent responsible for running process and monitor them inside the microVMs. | |
Runtimes | Runtimes to run functions inside the microVMs. | |
Gateway | Gateway is the entry point of the PolyXia platform. It is responsible for the communication with the user and the orchestration of the different components. | |
Firepilot | Rust API to pilot Firecracker. |
The documentation is available at : https://polyxia-org.github.io/polyxia
Contributions to Polyxia are welcome! Please see the CONTRIBUTING.md file for more information.
The Polyxia project belongs to Polytech Montpellier and is released under the MIT license. Please see the LICENSE file for more information.